mk e wrote:
You should NEVER trust it for cyl-cyl variation. There are just too many factors that can cause differences to know what the numbers mean. If you're on a budget and want cyl-cyl info then plug reads are about the only option, low resolution but they can't be wrong.
I usually use EGT to point at which cylinder is having issues. But, the absolute numbers aren't useful on their own.
mk e wrote:
With a slight budget I think NBO2 sensors are going to work.....I just need to prove it. EGT is normally used to look at change with time.....but I'm pretty sure setting spark to min EGT at any given mixture should be peak efficiency but it seems the same should be true for MBT and I guess BSFC too.
I'll need to ponder the BSFC thought. Efficiency can be improved with EGR and/or lean mixture but both require both fuel and spark changes....if mixture is fixed it seems like you'll never find optimal but would have the min BSFC...or are you changing fuel and spark looking at BSFC and basically ignoring mixture?
I usually do a three-step process:
-First, VE is calibrated at stoichiometric using only O2 feedback. This establishes that the f/a ratio commanded should equal the f/a ratio the engine operates at. Spark is safely retarded here, and so efficiency will be bad. The entire VE surface is mapped this way.
-Second, the lean limit (or EGR limit) is established, making a 'lean limit' surface of target F/A ratios. The right way to do this is to look at CoV IMEP, but without cylinder pressure sensors I use hydrocarbon emissions. Without an emissions bench, I would probably have to look for a steep increase in BSFC and then back off a bit. Torque numbers aren't very valid because spark is retarded and fixed and MBT varies with F/A ratio or EGR %, so neither is BSFC, but a steep BSFC trend will show the onset of misfire, and hydrocarbons/CO will show that sooner and are unrelated to actual torque produced.
-Third, the spark is calibrated at the lean or EGR limit. Now, the torque numbers can be trusted and the final BSFC should be good. When the spark mapping is done, the data collected can be compiled into a BSFC map of the engine, which is really useful for analysis later.
I've never calibrated EGR before, but we're designing an EGR system now so that should be fun. I suspect the three-step process will turn into a five-step process - Spark should be calibrated stoich before introducing EGR, and spark air temp corrections need to be done without EGR since EGR changes air temp.
I might take your suggestion of using narrowband sensors per cylinder. Are you controlling their heater temperature?